- Google now measures Gemini AI usage by the computing power of your requests, not the number of requests.
- There are four US tiers — Free, AI Plus ($8/mo), AI Pro ($20/mo), and AI Ultra ($100 or $200/mo) — with Plus at 2x free limits, Pro at 4x, and Ultra 5x-20x higher than Pro.
- Usage resets on a rolling five-hour window plus a weekly limit, checkable in the app under Usage limits.
- Context windows scale by tier: 32K tokens (free), 128K (Plus), and 1 million (Pro and Ultra).
What Happened
Google has revamped how it meters usage across its Gemini AI apps, measuring by the computing power a request demands rather than by request count, according to a July 18, 2026 guide from Wired. If you’ve suddenly hit a limit and been told to wait before prompting again, the new metering is likely why.
Why It Matters
The change makes sense from Google’s side — it charges by what a request actually costs its data centers — but it’s vaguer for users, who can no longer rely on a rule like “five image generations a day.” Google’s support docs add that “access is subject to change or may be limited based on testing, experimentation or availability,” which in practice means some days may allow more prompting than others.
Technical Details
Two main factors set how much you can use: your plan, and the complexity and length of your prompts (a weather query versus coding a mini-app). The specific model matters too — models like Gemini 3.5 Flash count differently — as do “thinking” levels (Standard, Extended, Deep Think). The paid US tiers are AI Plus ($8/month), AI Pro ($20/month), and AI Ultra ($100 or $200/month); Google describes free limits only as “standard,” then grants 2x on Plus, 4x on Pro, and 5x-20x on Ultra depending on payment level. Context windows scale from 32K tokens on free (about 24,000 words) to 128K on Plus (about 96,000 words) to 1 million on Pro and Ultra (about 750,000 words). To check your status, open Usage limits (the cog icon on web, or the menu then cog on mobile); a rolling bar resets every five hours and a second bar resets weekly.
Who’s Affected
Every Gemini user is affected, but heavy users of resource-intensive features like video generation will notice first, and Google’s docs say free users may be limited first when it needs to manage capacity. Hitting a limit on a paid plan demotes you to the most basic model until the next reset, rather than cutting you off entirely.
What’s Next
Google says limits may change without notice due to capacity constraints, so the practical takeaway is to monitor the in-app Usage limits screen rather than assume fixed quotas. The shift toward compute-based metering mirrors how AI providers are increasingly pricing by real resource cost — a trend likely to spread as models get more capable and expensive to run.
